public final Class OperandResolver

Implements (some perhaps not well known) Excel functionality to select a single cell from an area depending on the coordinates of the calling cell. Here is an example demonstrating both selection from a single row area and a single column area in the same formula.

The same concept is extended to references across sheets, such that even multi-row, multi-column areas can be useful.

Of course with carefully (or carelessly) chosen parameters, cyclic references can occur and hence this method can throw a 'circular reference' EvaluationException. Note that this method does not attempt to detect cycles. Every cell in the specified Area ae has already been evaluated prior to this method call. Any cell (or cells) part of ae that would incur a cyclic reference error if selected by this method, will already have the value ErrorEval.CIRCULAR_REF_ERROR upon entry to this method. It is assumed logic exists elsewhere to produce this behaviour.

Returns:ValueEval

whatever the selected cell's evaluated value is. Never null. Never ErrorEval.

Exceptions

EvaluationException:

if there is a problem with indexing into the area, or if the evaluated cell has an error.

Returns:Boolean

null to represent blank values

Exceptions

EvaluationException:

if ve is an ErrorEval, or if a string value cannot be converted

Applies some conversion rules if the supplied value is not already a number. Note - BlankEval is converted to NumberEval#ZERO.

Parameters

ev:ValueEval

must be a NumberEval, StringEval, BoolEval or BlankEval

Returns:double

actual, parsed or interpreted double value (respectively).

Exceptions

EvaluationException:

if a StringEval is supplied and cannot be parsed as a double (See parseDouble() for allowable formats).

IllegalStateException:

if the supplied parameter is not NumberEval, StringEval, BoolEval or BlankEval

Applies some conversion rules if the supplied value is not already an integer.
Value is first coerced to a double ( See coerceValueToDouble() ). Note - BlankEval is converted to 0.

Excel typically converts doubles to integers by truncating toward negative infinity.
The equivalent java code is:
  return (int)Math.floor(d);
not:
  return (int)d; // wrong - rounds toward zero

Parameters

ve:ValueEval

must be a NumberEval, StringEval, BoolEval, or BlankEval

Returns:String

the converted string value. never null

Retrieves a single value from an area evaluation utilizing the 2D indices of the cell within its own area reference to index the value in the area evaluation.

Parameters

ae:AreaEval

area reference after evaluation

cell:EvaluationCell

the source cell of the formula that contains its 2D indices

Returns:ValueEval

a NumberEval, StringEval, BoolEval or BlankEval. or ErrorEval Never null.

Retrieves a single value from a variety of different argument types according to standard Excel rules. Does not perform any type conversion.

Parameters

arg:ValueEval

the evaluated argument as passed to the function or operator.

srcCellRow:int

used when arg is a single column AreaRef

srcCellCol:int

used when arg is a single row AreaRef

Returns:ValueEval

a NumberEval, StringEval, BoolEval or BlankEval. Never null or ErrorEval.

Exceptions

EvaluationException:

if srcCellRow or srcCellCol do not properly index into an AreaEval. If the actual value retrieved is an ErrorEval, a corresponding EvaluationException is thrown.

Converts a string to a double using standard rules that Excel would use.
Tolerates leading and trailing spaces,

Doesn't support currency prefixes, commas, percentage signs or arithmetic operations strings. Some examples:
" 123 " -> 123.0
".123" -> 0.123
"1E4" -> 1000
"-123" -> -123.0
These not supported yet:
" $ 1,000.00 " -> 1000.0
"$1.25E4" -> 12500.0
"5**2" -> 500
"250%" -> 2.5

Returns:Double

null if the specified text cannot be parsed as a number